NMR Study of the Association of Propofol with Nonionic Surfactants

The general anesthetic 2,6-diisopropylphenol (propofol) is very poorly soluble in water and is normally administered in the form of an emulsion...
3 downloads 0 Views 95KB Size
Download PDF
2088

Langmuir 2003, 19, 2088-2095

NMR Study of the Association of Propofol with Nonionic Surfactants Konstantin I. Momot,† Philip W. Kuchel,*,† Bogdan E. Chapman,† Peter Deo,‡ and Darryl Whittaker‡ School of Molecular and Microbial Biosciences, University of Sydney, Sydney, NSW 2006, Australia, and DBL Australia Pty. Ltd., Level 21, 390 St. Kilda Road, Melbourne, Vic 3004, Australia Received October 21, 2002. In Final Form: December 17, 2002 The general anesthetic 2,6-diisopropylphenol (propofol) is very poorly soluble in water and is normally administered in the form of an emulsion. We demonstrated that several commercially available nonionic surfactants (Tween 80, Cremophor EL, Poloxamer 188, Poloxamer 407, Solutol HS15, and Vitamin E TPGS) render propofol soluble with a specific solubilization capacity of at least 0.1 g/g. The room-temperature stability of the solutions appeared to be limited only by the chemical stability of the compounds involved. The association between propofol and the surfactants was investigated by various NMR approaches, including measurements of diffusion coefficients, 1H longitudinal relaxation times, and the magnitude of intermolecular nuclear Overhauser effects. The results were consistent with the micellar solubilization mechanism of propofol by the surfactants (unimer solubilization in the case of Poloxamer 188). The 1H longitudinal relaxation and diffusion behavior of propofol were monoexponential in each case. Solubilization caused a considerable shortening of propofol’s proton T1’s. The values of the diffusion coefficient of propofol were several percent higher than those of surfactants. This was explained by the partitioning of propofol between swollen micelles and the aqueous solution. Diffusion measurements also revealed the presence of a rapidly diffusing ethylene oxide population in surfactant solutions, which is consistent with free poly(ethylene oxide) (PEO) known to be present in commercially produced surfactants. The free PEO blocks exhibited molecular association with the extramicellar propofol.

1. Introduction Propofol (2,6-diisopropylphenol) is a commonly used general anesthetic. In clinical practice it is administered intravenously. It is dispensed as an isotonic oil-in-water emulsion at a concentration of 1% (w/v) that is formulated with glycerol, soya oil, sodium hydroxide, egg lecithin, and disodium EDTA and has a pH of 6-8.5.1,2 The goal of this work was to investigate the association between propofol and six nonionic polymeric surfactants: Poloxamer 407 (F127), Poloxamer 188 (F68), Tween 80, Solutol HS15, Cremophor EL, and Vitamin E TPGS. The principal components of Poloxamers are triblock copolymers of the general form EOn1POmEOn2, where EO ) ethylene oxide and PO ) propylene oxide. The average composition {n1, m, n2} is {97, 69, 97} to {100, 65, 100} for F127 and {76, 29, 76} for F68; their average molecular weights are 12600 and 8400, respectively.3-7 Cremophor EL (average MW ∼ 2500) is polyethoxylated castor oil, a complex mixture of surfactants with the main component being poly(ethylene glycol)(35) glycerol triricinoleate.8 The * To whom correspondence should be addressed. E-mail address: [email protected]. † University of Sydney. ‡ DBL Australia Pty. Ltd. (1) Reynolds, J. E. F., Ed. Martindale: The Extra Pharmacopoeia, 32nd ed.; The Pharmaceutical Press: London, 1999. (2) Luckenbach, R., Ed. Beilsteins Handbuch der Organischen Chemie, 4th ed.; Springer-Verlag: Berlin, 1980; entry for propofol is 6 IV 3435. (3) Couderc, S.; Li, Y.; Bloor, D. M.; Holzwarth, J. F.; Wyn-Jones, E. Langmuir 2001, 17, 4818-4824. (4) Liu, T.; Nace, V. M.; Chu, B. Langmuir 1999, 15, 3109-3117. (5) Holmqvist, P.; Alexandridis, P.; Lindman, B. J. Phys. Chem. B 1998, 102, 1149-1158. (6) Alexandridis, P.; Hatton, T. A. Colloids Surf., A 1995, 96, 1-46. (7) Alexandridis, P.; Hatton, T. A. Macromolecules 1994, 27, 24142425.

principal components of the other surfactants are as follows: Solutol HS15, poly(ethylene glycol)(15) 12-hydroxystearate (MW ∼ 960); Vitamin E TPGS, R-tocopheryl poly(ethylene glycol)(22) succinate (MW ∼ 1500); Tween 80 (Crillet 4 HP, Polysorbate 80), tris-poly(ethylene glycol)(20) sorbitan monooleate (MW ∼ 1300).9,10 The principal criterion for the choice of surfactants for this study was their biocompatibility and pharmacological properties, briefly discussed below in this section. All of the surfactants contain free poly(ethylene oxide) (PEO), which is often added purposefully to facilitate the industrial handling of the compounds. Of the above list, Poloxamers are probably the most studied and the best understood, and have been extensively reviewed in the literature.6 Poloxamers 407 and 188 are also available under the names Lutrol F127 (Pluronic F127) and Lutrol F68 (Pluronic F68), respectively. Lutrol is the clinical grade of Poloxamers, extensively used in the pharmaceutical industry. The PEO and the PPO blocks are usually thought of as hydrophilic and hydrophobic, respectively.11 Poloxamers are thus amphiphilic block copolymer surfactants with relatively low polydispersity. Variation of the length of each of the blocks enables the modulation of the copolymer properties (such as micellization) in a wide range of their respective values. In this work, several NMR techniques were applied to investigate the degree and the nature of association between the studied surfactants and propofol. The prin(8) Priev, A.; Zalipsky, S.; Cohen, R.; Barenholz, Y. Langmuir 2002, 18, 612-617. (9) O’Neil, M. J., Ed. The Merck Index, 13th ed.; Merck: New Jersey, 2001. (10) Schick, M. J., Ed. Nonionic Surfactants; Marcel Dekker Inc.: New York, 1967. (11) Paterson, I. F.; Chowdhry, B. Z.; Leharne, S. A. Langmuir 1999, 15, 6187-6194.

10.1021/la026722g CCC: $25.00 © 2003 American Chemical Society Published on Web 02/06/2003

Propofol and Nonionic Surfactants

cipal findings were the following. (1) Each of the six surfactants solubilized propofol to at least 1% (w/v) at the 10% (w/v) surfactant concentration. (2) In surfactant solutions, poly(ethylene oxide) exhibited two diffusion coefficients, “fast” (free PEO blocks) and “slow” (micellar PEO blocks bound to the hydrophobic tails). (3) Propofol exhibited a single diffusion coefficient in all cases; its value was consistent with propofol largely being associated with the surfactant molecules or micelles and partly residing in the extramicellar solution. (4) In each system, the proton relaxation times of the solubilized propofol were considerably shorter than those in a CDCl3 solution or in bulk propofol. (5) Although diffusion coefficient measurements are required to fully understand the molecular association in the studied systems, T1 measurements were used to provide “quick evidence” of molecular association between propofol and the surfactants. The motivation for this work was, in part, the investigation of alternative delivery vehicles for the intravenous administration of propofol. While the commercially available emulsion (Diprivan) provides for an efficient delivery of propofol, it also has disadvantages. Two of these are the presence of residual free propofol in the “water” phase, which is believed to lead to pain on intravenous injection,12 and the propensity of the emulsion to support bacterial growth.13 In recent years extensive research has been undertaken to develop novel and robust drug delivery vehicles providing for fewer clinical side effects.14-18 A number of nonionic surfactants appear promising in this regard.19-21 Although problems with clinical applications of some surfactants (most notably with Cremophor) have been pointed out,22-24 many are widely used in drug formulations. They are generally acknowledged to have low toxicity and are well tolerated clinically.25,26 Several compounds containing poly(ethylene oxide) (PEO) moieties have been shown to modulate or reverse multidrug resistance in animal and human cancer cells, with Solutol HS15 and Tween 80 being among the least toxic.27-29 Solutol HS15 has been shown to interact with plasma (12) Larsen, B.; Beerhalter, U.; Biedler, A.; Brandt, A.; Doege, F.; Brun, K.; Erdkonig, R.; Larsen, R. Anaesthetist 2001, 50, 842-845. (13) Wachowski, I.; Jolly, D. T.; Hrazdil, J.; Galbraith, J. C.; Greacen, M.; Clanachan, A. S. Anesth. Analg. 1999, 88, 209-212. (14) Momot, K. I.; Whittaker, D.; Kuchel, P. W. Novel Drug Delivery Vehicles: NMR Approach. In The 43rd Experimental NMR Conference, Book of Abstracts; Pacific Grove, CA, 2002. (15) Riley, T.; Stolnik, S.; Heald, C. R.; Xiong, C. D.; Garnett, M. C.; Illum, L.; Davis, S. S.; Purkiss, S. C.; Barlow, R. J.; Gellert, P. R. Langmuir 2001, 17, 3168-3174. (16) von Corswant, C.; Thoren, P. E. G. Langmuir 1999, 15, 37103717. (17) Drummond, C. J.; Fong, C. Curr. Opin. Colloid Interface Sci. 2000, 4, 449-456. (18) Shah, J. C.; Sadhale, Y.; Chilukuri, D. M. Adv. Drug Delivery Rev. 2001, 47, 229-250. (19) Torchilin, V. P. J. Controlled Release 2001, 73, 137-172. (20) Paavola, A.; Yliruusi, J.; Rosenberg, P. J. Controlled Release 1998, 52, 169-178. (21) Hagan, S. A.; Coombes, A. G. A.; Garnett, M. C.; Dunn, S. E.; Davies, M. C.; Illum, L.; Davis, S. S. Langmuir 1996, 12, 2153-2161. (22) Bettschart-Wolfensberger, R.; Semder, A.; Alibhai, H.; Demuth, D.; Shojaee Aliabadi, F.; Clarke, K. W. J. Vet. Med. A 2000, 47, 341350. (23) Rowinsky, E. K.; Donehower, R. C. New Engl. J. Med. 1995, 332, 1004-1014. (24) Teague, W. R. New Zealand Vet. J. 1986, 34, 104-105. (25) Lee, H.-K.; Jin, J.-Y.; Cho, H. Anesthetic Composition for Intravenous Injection Comprising Propofol. International patent application WO 00/78301, 2000. (26) Hong, J. W.; Ryoo, H. K.; Lee, K. H.; Chi, S. C.; Park, E. S. Solubilization of Propofol in Nonionic Surfactant Systems. In Proceedings of the 29th Annual Meeting; Controlled Release Society: Seoul, 2002. (27) Woodcock, D. M.; Linsenmeyer, M. E.; Chojnowski, G.; Kriegler, A. B.; Nink, V.; Webster, L. K.; Sawyer, W. H. Br. J. Cancer 1992, 66, 62-68. (28) Coon, J. S.; Knudson, W.; Clodfelter, K.; Lu, B.; Weinstein, R. S. Cancer Res. 1991, 51, 897-902.

Langmuir, Vol. 19, No. 6, 2003 2089

lipoproteins,30 and Poloxamers have been shown to interact with brain microvessel endothelial cells.31 Poloxamers have been shown to enhance the delivery of drugs and ATP to cells31,32 as well as to affect the distribution of the delivered drug inside the target cell.33 Targeted drug delivery mediated by Poloxamers conjugated with an antibody has been attempted.32 Poly(ethylene glycol) coating has been found to increase the bloodstream circulation lifetime of liposomes.34 All of these features make PEO-containing polymer surfactants particularly interesting as potential components of intravenous drug delivery vehicles. The range of surfactants studied in this work was not meant to be comprehensive, with Triton and Brij series being two of the notable omissions. However, the methodology presented here is independent of the specific surfactant. The development of a detailed practical example of the application of NMR to a pharmacologically relevant micellar system formed the second part of our motivation. The NMR results described here are relatively easy to interpret, and the experiments used could provide a basis for a methodology used for the in vitro evaluation of novel drug delivery systems. 2. Experimental Section Materials. Surfactants were obtained from the following sources: Tween 80 (Crillet 4 HP, CAS 9005-65-6) from Croda Chemicals (Australia); Cremophor EL (61791-12-6), Poloxamer 188 (106392-12-5), Poloxamer 407 (106392-12-5), and Solutol HS15 (61909-81-7) from BASF (Ludwigshafen, Germany); Vitamin E TPGS (9002-96-4) from Eastman Chemicals (Kingsport, TN). The certificates of analysis of the batches used showed that they typically consist of approximately equal amounts of the principal amphiphilic compound and free poly(ethylene glycol), with small (∼5%) quantities of impurities. Propofol (CAS 207854-8) was obtained from Archimica SpA (Varese, Italy). CDCl3 was purchased from Aldrich (Milwaukee, WI); CCl4 (used for magnetic susceptibility matching, spectroscopic grade), from AJAX Chemicals (Auburn, NSW, Australia); D2O, from ANSTO (Lucas Heights, NSW, Australia). All chemicals were used as received. Water was obtained from a Milli-Q reverse-osmosis apparatus (Millipore, Bedford, MA). Surfactant Solutions. The typical surfactant concentration in the studied systems was 10% (w/v). The solutions were prepared by repeated vortexing and incubation at 35-40 °C over at least 24 h. Propofol was loaded into the D2O/surfactant systems using the same cycle. For the solutions of Poloxamers, the preparation procedure included centrifugation at 380g to facilitate the dissolution of the surfactant. Cloud point and gelation temperatures were estimated visually following the equilibration of samples in a water bath. NMR Spectrometer. A Bruker (Karlsruhe, Germany) DRX400 spectrometer with an Oxford Instruments (Oxford, U.K.), 9.4 T, wide-bore magnet equipped with a 1000 G cm-1 z-only, actively shielded diffusion gradient probe was used in the experiments. The general setup of the spectrometer has been described elsewhere.35-39 The length of the linear-gradient region was 1 cm in the z direction. The probe was equipped with a number of inserts, of which the 10-mm 1H-only and the 5-mm inner-31P, outer-1H inserts were used for the 1H diffusion measurements. 1H 90°-pulse durations were ∼20 and 40 µs, (29) Batrakova, E.; Lee, S.; Li, S.; Venne, A.; Alakhov, V.; Kabanov, A. Pharm. Res. 1999, 16, 1373-1379. (30) Woodburn, K.; Sykes, E.; Kessel, D. Int. J. Biochem. Cell Biol. 1995, 27, 693-699. (31) Miller, D. W.; Batrakova, E. V.; Waltner, T. O.; Alakhov, V. Y.; Kabanov, A. V. Bioconjugate Chem. 1997, 8, 649-657. (32) Kabanov, A. V.; Batrakova, E. V.; Meliknubarov, N. S.; Fedoseev, N. A.; Dorobnich, T. Y.; Alakhov, V. Y.; Chekhonin, V. P.; Nazarova, I. R.; Kabanov, V. A. J. Controlled Release 1992, 22, 141-157. (33) Rapoport, N.; Marin, A.; Luo, Y.; Prestwich, G. D.; Muniruzzaman, M. J. Pharm. Sci. 2002, 91, 157-170. (34) Torchilin, V. P.; Papisov, M. I. J. Liposome Res. 1994, 4, 725739.

2090

Langmuir, Vol. 19, No. 6, 2003

respectively. Most of the inversion recovery measurements, NOESY, and reference 1D measurements were performed using a 5-mm Bruker TXI probe (typical 1H 90°-pulse length, 8 µs). Sample temperature was controlled, where applicable, by air flow at 400 L h-1. Temperature was calibrated separately for each probe using a capillary containing methanol (low-T) or ethylene glycol (high-T).40,41 NMR Samples. To ensure that the sample in diffusion measurements was completely contained within the probe’s constant-gradient region, it was constrained to the length 8-9 mm. One of the following three setups was used when measuring the diffusion coefficients: (1) 10-mm 1H-only insert, the sample (∼0.7 mL) was placed into a cylindrical Wilmad microcell, bubbles were removed by light tapping, and the microcell was placed inside a 10-mm o.d. NMR tube containing 2.5 mL of CCl4 for magnetic susceptibility matching;37,39 (2) 10-mm 1H-only insert, the sample (0.7 mL) was placed into a 9-mm o.d. flat-bottom NMR tube, a 13-mm long Teflon susceptibility plug was placed above the sample, and the tube was inserted into a 10-mm o.d. NMR tube containing 1.5 mL of CCl4 for magnetic susceptibility matching; (3) 5-mm inner-31P, outer-1H insert, the sample was placed into a 5-mm o.d. Shigemi tube (BMS-005B, susceptibilitymatched for D2O), the sample’s length was restricted by the susceptibility-matching rod. The typical shimming line width was 8-10 Hz in setups 1 and 2 and 3-5 Hz in setup 3. When the TXI probe was used, either the standard Wilmad 528-PP NMR tubes or the susceptibility-matched Shigemi tubes were used; the residual line width was ∼3 Hz in both cases. NMR spectra, diffusion coefficients, and longitudinal relaxation times were measured at 23 °C. No chemical shift standard was added, to obviate the possibility of it interacting with the micelles. Chemical shift referencing was done on the HDO peak (set to 4.8 ppm). A small amount of CDCl3 (∼15% v/v) was added to bulk propofol samples for the lock signal. NMR Measurements. A convection-compensating PGSE pulse sequence42 was used for most diffusion measurements. Some of the room-temperature measurements were carried out using the noncompensated PGSE sequence. Typically, 32 linearly incremented values of g were used in a diffusion measurement. Trapezoidal gradient pulses were used (ramped in 10 steps, ramp duration 0.1 or 0.5 ms; the exact value had no identifiable effect on the quality of the measured data). The absence of convection effects was established by comparing the results of measurements carried out with different values of the diffusion time (e.g. 8 and 20 ms). Longitudinal relaxation times, T1, were measured using a standard inversion recovery pulse sequence. NOESY spectra were acquired using either the standard43 or gradient-selected44 phase-sensitive pulse sequence. Data processing is described in detail elsewhere.35,45 The spectra of bulk propofol were recorded with an appropriately detuned Bruker TXI probe. The absence of radiation damping effects was checked by using a modified inversion-recovery pulse sequence, π/2-τ-π/2. In the determination of relaxation times and diffusion coefficients of micellar propofol, particular attention was paid to baseline correction of the spectra in order to avoid distortions of the measured intensities of the inherently small propofol peaks. In most cases, (35) Piton, M. C.; Lennon, A. J.; Chapman, B. E.; Kuchel, P. W. J. Colloid Interface Sci. 1994, 166, 437-443. (36) Kuchel, P. W.; Chapman, B. E. J. Magn. Reson. 1991, 94, 574580. (37) Waldeck, A. R.; Kuchel, P. W.; Lennon, A. J.; Chapman, B. E. Prog. Nucl. Magn. Reson. Spectrosc. 1997, 30, 39-68. (38) Lennon, A. J.; Scott, N. R.; Chapman, B. E.; Kuchel, P. W. Biophys. J. 1994, 67, 2096-2109. (39) Dingley, A. J.; Mackay, J. P.; Chapman, B. E.; Morris, M. B.; Kuchel, P. W. J. Biomol. Nucl. Magn. Reson. 1995, 6, 321-328. (40) Ammann, C.; Meier, P.; Merbach, A. E. J. Magn. Reson. 1982, 46, 319-321. (41) Momot, K. I.; Walker, F. A. J. Phys. Chem. A 1997, 101, 92079216. (42) Sørland, G. H.; Seland, J. G.; Krane, J.; Anthonsen, H. W. J. Magn. Reson. 2000, 142, 323-325. (43) Bodenhausen, G.; Kogler, H.; Ernst, R. R. J. Magn. Reson. 1984, 58, 370-388. (44) Wagner, R.; Berger, S. J. Magn. Reson. A 1996, 123, 119-121. (45) Momot, K. I.; Kuchel, P. W. Drug Delivery Vehicles from an NMR Perspective. Concepts Magn. Reson., to be submitted for publication, 2003.

Momot et al.

Figure 1. Proton NMR spectra: (A) Tween 80; (B) Vitamin E TPGS; (C) Poloxamer 407 (all at 10% (w/v) concentration in D2O-saline). In parts A and B, the biggest peak (3.7 ppm) is a superposition of -CH2- peaks from the micellized and the free PEO blocks. In part C, it is a superposition of the same peaks from the same two PEO populations as well as -CH- peaks from both the micellar and the free PPO blocks. The peak near 0.9 ppm in parts A and B is from aliphatic CH3 groups. The peak near 1.3 ppm in part A belongs to aliphatic (non-EO) -CH2- groups. The peak near 1.1 ppm in part C is from PPO CH3 groups. The smaller peaks in the aliphatic region in parts A and B belong to various moieties of the hydrophobic sorbitan and Vitamin E tails, respectively. polynomial baseline correction (second order) was sufficient. In some cases, cubic-spline correction was used.

3. Results Proton NMR spectra of all the surfactant solutions in D2O were similar in that in each of these the largest peak was that near 3.7 ppm corresponding to the CH2 groups of poly(ethylene oxide) (PEO) and poly(isopropylene oxide) (PPO), as well as the CH protons of PPO. This peak was inhomogeneously broadened in all samples, and its line shape depended on the presence of propofol in the system. Other prominent peaks, common to all or several of the compounds, were near 0.9 ppm from CH3 groups (1.2 ppm for PPO CH3 groups) and (except for the Poloxamers) near 1.3 ppm from aliphatic CH2 groups. Representative spectra of three of the surfactants are shown in Figure 1. Propofol Solubilization: Initial Observations. The structure of propofol and the 1H NMR spectrum of its solution in CDCl3 are shown in Figure 2. T1 values of its protons in the CDCl3 solution ranged from 1.7 s for the methyl groups to 3.6 s for the aromatic proton in the paraposition to the OH group. T1’s in bulk propofol were approximately half these values.

Propofol and Nonionic Surfactants

Langmuir, Vol. 19, No. 6, 2003 2091

Table 1. Chemical Shifts (ppm) of Several 1H NMR Peaks of Propofol-Free and Propofol-Containing Surfactant Solutions in D2O-Salinea 10% (w/v) S/D2O-saline surfactant

PEO/PPO CH2b

F68 F127 Sol15 T80 CEL VE

3.722 3.724 3.721 3.720 3.722 3.712

a

1% (w/v) P/10% (w/v) S/D2O-saline

non-EO CH2c

PPO or non-PPO CH3c

PEO/PPO CH2b

1.325 1.323 1.321 1.280

1.188d 1.170d 0.928e 0.915e 0.920e 0.881e

3.720 3.710 3.728 3.708 3.707 3.689

non-EO CH2c

PPO or non-PPO CH3c

1.273 1.256 1.250 1.270

1.169 and 0.994d 1.050d 0.868e 0.857e 0.851e 0.853e

The HDO peak was used as a reference (set nominally to 4.8 ppm). b “Hydrophilic”. c “Hydrophobic”.

d

PPO. e Aliphatic non-PPO.

Figure 2. Chemical structure of propofol and its 1H NMR spectrum in a CDCl3 solution. Assignments (400 MHz, δ): 7.15 (d, J ) 7.6 Hz, 2H, Ar H3 and H5), 7.00 (t, J ) 7.6 Hz, 1H, Ar H4), 4.89 (br s, 1H, OH), 3.25 (septet, J ) 6.9 Hz, 2H, -CH(CH3)2), 1.36 (d, J ) 6.9 Hz, 12H, CH3).

All of the surfactant solutions solubilized propofol. At room temperature, a 10% (w/v) solution of each surfactant was capable of solubilizing propofol to at least 1% (w/v) concentration, which corresponded to a specific solubilization capacity of 10%. The solutions were clear, had pale to intense yellow color, and showed no opalescence. They did not appear to “age”; all of them were chemically and physically stable for at least 2 months at room temperature and at least 1 year at 4 °C. The F68 solution of propofol formed a dark-yellow precipitate at 4 °C; however, a brief room-temperature vortexing restored the system and its NMR behavior to the original state. The other solutions remained homogeneous liquids when cooled to 4 °C. We estimated the clouding point or gelation temperature of each of the six systems containing 1% (w/w) propofol and 10% (w/w) surfactant in D2O-saline (P/S/D2O). The results were as follows: P/Solutol HS15/D2O-saline, cloud point > 40 °C; P/Vitamin E TPGS/D2O-saline, liquid crystalline transition (gelation) < 42 °C; P/Tween 80/D2O-saline, cloud point > 40 °C; P/Cremophor EL/D2O-saline, cloud point > 50 °C; P/either Poloxamer/D2O-saline, gelation > 50 °C. 1 H NMR peaks from hydrophobic groups shifted to low frequency in the presence of propofol. Table 1 lists the chemical shifts of selected peaks in the 10% (w/v) surfactant/D2O-saline and 1% (w/v) propofol/10% (w/v) surfactant/D2O-saline systems (S/D2O-saline and P/S/D2Osaline, respectively). The average propofol-induced, lowfrequency shift for the “hydrophilic” PEO CH2 peaks was 0.01 ( 0.01 ppm, while for the “hydrophobic” ones it was 0.07 ( 0.05 ppm. In addition, propofol effected either a change of the line shape or a partial splitting of the PEO peak at 3.7 ppm. The extent of the changes depended on the surfactant.

Figure 3. (A) Proton NMR spectrum of the system 1% (w/v) propofol/10% (w/v) Solutol HS15/D2O-saline (P/Sol/D2O-saline). This spectrum is representative of other P/surfactant/D2O-saline systems studied. The solid arrows (peaks a and b) show propofol peaks which were used for the determination of its diffusion coefficient. The dashed arrows (peaks c and d) show other propofol peaks. Peak e belongs to PEO-HS ester; it was not used because of the overlap with the propofol peak. (B and C) Detail of the PEO line shape in propofol/Solutol HS15/D2Osaline and Solutol HS15/D2O-saline, respectively. Propofol induced a partial splitting of the PEO peak.

Propofol + Solutol HS15. The micellar diffusion coefficient of PEO-HS ester in a propofol-free Solutol solution [9.15% (w/w)] was 1.9 × 10-11 m2 s-1. As an example of propofol solubilization, the system consisting of 1% (w/v) propofol and 10% (w/v) Solutol HS15 in D2Osaline (P/Sol/D2O-saline) was studied in detail. Its proton NMR spectrum is shown in Figure 3A and B. The PEO peak near 3.7 ppm experienced a partial splitting in the presence of propofol, which is evident from a comparison of parts B and C of Figure 3. The two peaks from aromatic protons near 7 ppm were used as diffusion and relaxation indicators of propofol. The CH3 peak at 0.87 ppm and the slowly diffusing component of the PEO peak at 3.73 ppm were used as indicators for the surfactant micelles.

2092

Langmuir, Vol. 19, No. 6, 2003

Table 2.

Momot et al.

1H

NMR Longitudinal Relaxation Times and Diffusion Coefficients in the System 1% (w/v) Propofol/10% (w/v) Solutol HS15/D2O-Salinea

1H-Determined

δ (ppm)

compd

T1 (ms)

6.93 6.74 3.73

P P Sol15

570 570 590

0.87

Sol15

560

D

(m2

s-1)

1.42 × 10-11 1.47 × 10-11 1.25 × 10-11 (28%)b 1.39 × 10-10 (72%)c 1.29 × 10-11

The standard deviation of the reported T1 values was (30 ms; that of the diffusion coefficient values was (0.07 × 10-11 m2 s-1. b 12-Hydroxystearate-bound micellar PEO. c Free PEO. a

Aliphatic peaks at 1.27 ppm (PEO-HS ester) and 1.19 ppm (propofol) overlapped, and only a single T1 and D could be determined for these peaks. The T1 and D values determined from the nonoverlapping peaks are presented in Table 2. The peak at 1.27/1.19 ppm showed monoexponential longitudinal relaxation in all cases, and the value of D obtained from it (1.42 × 10-11 m2 s-1) was consistent with those listed in Table 2. The observed value of the diffusion coefficient of propofol was (1.44 ( 0.1) × 10-11 m2 s-1. The diffusion coefficient of the surfactant (polyethoxylated 12-hydroxystearate) in the propofol-containing solution was (1.27 ( 0.1) × 10-11 m2 s-1. The NOESY spectrum of the P/Sol/D2O-saline system revealed strong negative intermolecular NOEs between the aromatic propofol peaks and the Solutol HS15 peaks at 3.73 and 0.87 ppm. Propofol and Other Surfactants. Table 3 shows propofol and surfactant diffusion coefficients in each of the propofol/surfactant/D2O-saline systems. In each of these, 1H PFG NMR diffusion measurements showed the presence of two PEO populations, rapidly and slowly diffusing. In addition to the “fast” PEO population, in the two Poloxamers the polypropyleneoxide (PPO) peak near 1.18 ppm also exhibited two separate diffusion coefficients. In each of the six systems, the Stejskal-Tanner plots of propofol signals were linear. The effective diffusion coefficients were slightly higher than the “slow” diffusion coefficients of the respective surfactants. The values of all the diffusion coefficients were independent of the diffusion interval ∆ in the range studied. From a comparison of several measurement methods, the relative standard deviation of the measured diffusion coefficients was estimated to be 5%. A study of the concentration dependence of the diffusion coefficients of the surfactants and propofol was not made. However, measurements of the Solutol HS15/D2O-saline system revealed that the dependence of the micellar D of Solutol on its concentration in D2O in the range 1-5% (w/v) was significant, that is, exceeded the standard deviation of the measurement. This means that the values of the diffusion coefficients reported here should be treated as concentration-specific. Table 4 lists the T1 values of the two aromatic protons of propofol in three of the studied systems. The standard deviation of the reported values is conservatively estimated at (30 ms. Propofol T1 values in the other three systems were less precise but were consistent with the three former systems, that is, near 500 ms. NOESY spectra were also recorded for the propofol/ Tween 80/D2O-saline system at 22 and 35 °C. The sorbitan-propofol intermolecular NOEs were negative at 22 °C and weakly positive at 35 °C. The PEO-propofol NOEs were weakly negative and weakly positive, respectively.

4. Discussion Although the phase diagrams of surfactant systems can be very complex,46,47 in this work we focused only on the simple disordered micellar and unimer phases which exist in dilute solutions of surfactants. All of the studied surfactants, with the exception of Poloxamer 188, have critical micellization concentrations well below 1% (w/v) at biologically relevant temperatures. F68, on the other hand, exists in a 10% (w/v) water solution in the unimer form up to 33 °C.7 The range of surfactants investigated thus permitted a comparison of the micellar and the unimer solubilization of propofol. As can be expected of high-molecular weight compounds, the room-temperature values of the diffusion coefficients of each of the surfactants in their water solutions were of the order of 10-11 m2 s-1 or smaller. The proton longitudinal relaxation times of Solutol HS15 were measured both above and below the cmc. The latter values were slightly longer than the former. This is consistent with results reported for surfactants of similar MW48 and with the assumption that the rotational reorientation time of a single surfactant molecule lies near the crossover from the extreme-narrowing limit to the slow-motion limit. The proton T1’s in each surfactant did not exceed 1 s either above or below the cmc and were in the range 350-700 ms when the surfactants were in micellar form. Literature values of the cmc for each of the studied surfactants are listed in Table 5. Except for Poloxamer 188 (F68), all of the compounds are known to be in the micellar form in water solution in the temperature range 20-40 °C and concentration range 1-15% (w/v).5,49,50 The presence of electrolytes is known to affect the phase diagrams of surfactant/water systems.47 However, the effect is not significant at the physiological saline concentration (0.9% ) 154 mM NaCl). This was confirmed by our measurements carried out on pure-D2O solutions; the results were essentially the same as those for the D2O-saline solutions discussed below. Each surfactant contains a significant quantity of free PEO whose protons have the same chemical shift as those of the micellar surfactant PEO. Therefore, in diffusion experiments the single PEO 1H NMR peak corresponded to two nonexchanging populations, micellized and free PEO, and exhibited two diffusion coefficients. Propofol Solubilization: Initial Observations. In contrast to the macromolecular surfactants, propofol is a small molecule. It is a dark-yellow liquid almost insoluble in water (∼150 mg/L) but soluble in many organic solvents.2,51 A comparison of the spectrum in Figure 2 with those of the surfactants (cf. Figure 1) revealed that the two aromatic proton multiplets near 7 ppm were always well-separated from all surfactant peaks and were therefore suitable as indicators of propofol behavior in surfactant solutions. The other peaks overlapped with surfactant peaks either partially or fully. Hence, D and T1 values measured from the latter were generally less reliable than those measured from the former. (46) Alexandridis, P.; Olsson, U.; Lindman, B. Langmuir 1998, 14, 2627-2638. (47) Desai, P. R.; Jain, N. J.; Badahur, P. Colloids Surf., A 2002, 197, 19-26. (48) Yuan, H.-Z.; Cheng, G.-Z.; Zhao, S.; Miao, X.-J.; Yu, J.-Y.; Shen, L.-F.; Du, Y.-R. Langmuir 2000, 16, 3030-3035. (49) Wanka, G.; Hoffmann, H.; Ulbricht, W. Macromolecules 1994, 27, 4145-4159. (50) Eastman. Vitamin E TPGS NF: Properties and Applications; Kingsport, TN, 2000. Information also available on http://www.eastman.com/product_information/producthome.asp?product)1113. (51) Trapani, G.; Lopedota, A.; Franco, M.; Latrofa, A.; Liso, G. Int. J. Pharm. 1996, 139, 215-218.

Propofol and Nonionic Surfactants

Langmuir, Vol. 19, No. 6, 2003 2093

Table 3. Values of the Apparent Diffusion Coefficients of Propofol and Surfactants in the Respective 1% (w/v) Propofol/ 10% (w/v) Surfactant/D2O-Saline Systemsa D (m2 s-1)

a

surfactant

propofol

surfactant (slow component)

fast PEO component

fast PPO component

Sol15 T80 CEL VE F68 F127

1.44 × 10-11 1.64 × 10-11 1.47 × 10-11 1.20 × 10-11 7.10 × 10-12 2.65 × 10-12

1.27 × 10-11 1.42 × 10-11 1.40 × 10-11 1.04 × 10-11 5.43 × 10-12 6.85 × 10-13

1.39 × 10-10 3.72 × 10-11 1.06 × 10-10 3.59 × 10-11 2.03 × 10-11 2.05 × 10-11

n/a n/a n/a n/a 1.62 × 10-11 1.55 × 10-11

The relative standard deviation of the diffusion coefficient values was (5%.

Table 4. Longitudinal Relaxation Time (T1) of Aromatic Protons in Solubilized Propofola T1 (ms) surfactant Sol15 F68 F127 a

H3, H5

H4

570 630 560

570 710 610

The typical standard deviation of the T1 values was (30 ms.

Table 5. Literature Values of the Critical Micellization Concentration of the Studied Surfactantsa surfactant F68 F127 Sol15 T80 CEL VE

cmc (% w/v)

cmc (mM)

T (°C)

ref

15 7 4 0.7 0.008 0.02 0.0013 0.01 0.02

17.9 8.333 3.174 0.555 0.006 0.21 0.01 0.04 0.13

27 40 20 25 40 25 25 25 37

7 7 57 58 8 50

a Abbreviations: F68, Poloxamer 188 (Lutrol F68); F127, Poloxamer 407 (Lutrol F127); Sol15, Solutol HS15; T80, Tween80 (Crillet 4 HP, Polysorbate80); CEL, Cremophor EL; VE, Vitamin E TPGS.

The first indicators that the solubilization involved direct molecular association between propofol and the surfactants were the systematic low-frequency shift of 1H NMR signals from hydrophobic protons (Table 1) and the changes of the PEO line shape induced by propofol. The average low-frequency shift in the presence of 1% (w/v) propofol for the “hydrophilic” and “hydrophobic” peaks was 0.01 ( 0.01 and 0.07 ( 0.05 ppm, respectively. Although the possibility of a comparison of absolute shifts of such magnitude, when the referencing is done on the temperature- and pH-dependent HDO peak, is questionable, the relative shift of one group of protons versus the other is unambiguous. The hydrophobic non-EO CH2 and CH3 protons were thus shifted to low frequency considerably more than the relatively hydrophilic EO CH2 protons were. This is consistent with the former being in contact with domains possessing a less negative magnetic susceptibility than that of D2O, which is expected of an aromatic compound like propofol.52 Propofol + Solutol HS15. Diffusion studies of Solutol HS15 revealed that the ethylene oxide peak near 3.7 ppm exhibited two diffusion coefficients which differed roughly by about an order of magnitude and had population ratios of approximately 2(fast):1(slow). The two observed diffusion coefficients were attributed to the free PEO and the PEO blocks of the micellized surfactant, polyethoxylated 12-hydroxystearate. The diffusion coefficient of PEO(52) Doty, F. D.; Entzminger, G.; Yang, Y. A. Concepts Magn. Reson. 1998, 10, 133-156.

HS ester measured from the signal at 0.85 ppm was consistent with the “slow” PEO D to within 3% (see Table 2). T1 values of propofol indicate the local viscosity of propofol domains. In bulklike propofol droplets, propofol protons would have T1 values of approximately 1 s. Propofol molecules dispersed in a micellar core would experience a greater local viscosity and therefore have longer reorientation times and shorter T1 values. T1 values of propofol protons thus allow us to distinguish between bulklike propofol and propofol dispersed in the micellar core. (This can be compared to the hydration test in W/O microemulsions which has been used to distinguish between bulklike “active” water and water dissolved in oil.53) In the system propofol/Solutol HS15/D2O-saline, the T1 values of propofol protons are considerably shorter than those in bulk propofol, indicating that propofol domains inside the micelles do not possess a bulklike viscosity. On the other hand, a partial splitting of the PEO peak was present in this system (cf. Figure 3B and C). The diffusion coefficient of propofol-loaded PEO-HS micelles was smaller than the micellar D in the system Solutol HS15/D2O-saline (1.9 × 10-11 m2 s-1) by a factor of 1.5. This proves that the uptake of propofol increased the size of the surfactant micelles. The observed specific solubilization capacity of g10% is plausible for micellar solubilization.54,55 The observed diffusion coefficient of propofol (1.44 × 10-11 m2 s-1) was close to, but slightly higher than, the observed micellar diffusion coefficient of PEO-HS ester (1.27 × 10-11 m2 s-1). For a single system, this could be attributed to an experimental error. However, the results for other surfactants, presented in Table 3, revealed that this was a consistent phenomenon. We surmise that the explanation lies in the partitioning and fast chemical exchange of propofol molecules between the micelles and the extramicellar medium. The solubility of propofol in water (∼150 mg/L) is sufficiently high to account for the difference between the two diffusion coefficients. However, the presence of intermolecular NOEs between PEO and propofol indicates that the extramicellar propofol is likely to be associated with the free PEO blocks. This model is illustrated in Figure 4. The observed propofol diffusion coefficient is thus the weighted average of the diffusion coefficients of micelles and the free PEO.45 The respective values correspond to approximately 1% of propofol residing in the latter. This outcome is plausible considering that PEO blocks constitute the relatively hydrophilic head of the Solutol molecule, while the 12-hydroxystearic acid moiety forms the hydrophobic tail. Propofol and Other Surfactants. Free PEO was present in all of the studied surfactant solutions. There was a loose correlation between its diffusion coefficient (53) Wells, M. A. Biochemistry 1974, 13, 4937-4942. (54) Weiss, J.; McClements, D. J. Langmuir 2000, 16, 5879-5883. (55) Nagarajan, R.; Ganesh, K. J. Colloid Interface Sci. 1996, 184, 489-499.

2094

Langmuir, Vol. 19, No. 6, 2003

Momot et al. Table 6. Estimated Micellar Diameter and Rigid-Sphere Micellar Reorientation Time for the Studied Surfactantsa

Figure 4. Reaction scheme depicting the association of propofol with surfactant micelles and free PEO blocks in surfactant solutions. On the basis of the values of the observed diffusion coefficients shown in Table 3, approximately 99% of propofol in non-Poloxamer solutions is micellized. In Poloxamer solutions, PEO is replaced by PPO and the fraction of micellized propofol was estimated to be approximately 85%.

and the size of the PEO block in the surfactant molecule. The diffusion coefficient of HDO was in the range (1.11.5) × 10-9 m2 s-1 for all samples, meaning that the values of the effective hydrodynamic viscosities of the solutions were within 40% of each other. In the Poloxamer solutions, the two diffusion coefficients measured from the PPO peak near 1.18 ppm were interpreted as belonging to the free PPO and the micellar PPO populations, respectively. With the exception of the cases of the Poloxamers, the observed diffusion coefficients of propofol (Table 3) were very close to but a few percent higher than the values of the micellar diffusion coefficients of the respective surfactants. The explanation is the same as that in the case of Solutol HS15. Propofol molecules essentially resided in the micelles. However, ∼1% of the total propofol was associated with free PEO. The latter possessed a relatively large D value, which caused the observed diffusion coefficient of propofol to be slightly larger than that of the micelles. In the case of the Poloxamers, the more hydrophobic free PPO was also present in the solution. This further increased the apparent diffusion coefficient of propofol. The large difference between the apparent D of propofol and the micellar D of the surfactant can be misleading. Assuming that half of all PO is in the free-PPO form and that the affinity of propofol to free PPO is much greater than that to the free PEO, ∼15% of propofol needs to be associated with the free PPO in order to explain the apparent propofol D values in the two Poloxamer solutions. This would correspond to an extramicellar concentration of propofol of 0.15% (w/v), which significantly exceeds its solubility in water, 0.015% (w/w).9 This supports the hypothesis that the extramicellar propofol is associated with PEO or PPO blocks rather than present in the free form. Poloxamer 188 is a special case because it does not micellize under the conditions used in the present study. This was reflected in its “slow” D being almost an order of magnitude greater than that of F127, although the lengths of the unimers differ by only a factor of 1.4-1.5. This implies that micelles were indeed absent from the Poloxamer188/propofol/D2O-saline system, and the “slow” D was that of the unimer. The value of the observed diffusion coefficient of propofol in this system was consistent with the foregoing model; the measured values of the unimer diffusion coefficient; and the assumption that the majority of propofol molecules are associated with the Poloxamer188 unimers. Therefore, Poloxamer188 solubilized propofol despite not being able to form micelles under the investigated conditions. At the same time, the copolymeric unimer is a “minimicelle” in its own right, because it possesses a hydrophobic (PPO) and two hydrophilic (PEO) blocks. These blocks act as a core and a corona of the “minimicelle”, respectively. Micellar size and hard-sphere reorientation time can be formally estimated from the micellar diffusion coefficients using the Stokes-Einstein model as r ) kT/6πηD and τr ) ηV/kT, where k is Boltzmann’s constant, T is the

surfactant

d (nm)

τr (ms)

Sol15 T80 CEL VE F68 F127

15 ( 0.8 13 ( 0.7 13 ( 0.7 18 ( 0.9 35 ( 1.8 274 ( 14

3.8 ( 0.5 2.7 ( 0.4 2.8 ( 0.5 6.9 ( 1.0 49 ( 7 24000 ( 3500

a Obtained using the Stokes-Einstein model (see text). The viscosity of the solution was taken as η(D2O) ) 1.157 cP59 at T ) 23 °C. Rigid-sphere reorientation times greatly exceeded realistic values of molecular correlation times of both the surfactant and propofol (refer to Discussion). The nominal error intervals were calculated assuming that the relative standard deviation of the D values was 5%.

temperature, η is the solvent viscosity, τr is the molecular reorientation time, and r and V are the molecular hydrodynamic radius and volume, respectively. These estimates, assuming that the viscosity of the solution is that of D2O, 1.157 cP, are given in Table 6. However, they are not indicative of actual molecular reorientation times, as will be evident from the following discussion. Because there is no direct relationship between the micellar hydrodynamic radius and aggregation number,6 the values in Table 6 should not be used to estimate the latter. Consistent with the micellar solubilization of propofol is the shortening of its proton T1 values in surfactant solutions. It is clear that its uptake by surfactant micelles caused a lengthening of the molecular reorientation time of propofol; however, it was problematic to estimate by how much, on the basis of the available experimental information. The reorientation time of propofol in CDCl3 was estimated from propofol’s molecular size and solvent viscosity at ∼0.03 ns. Using the well-known expression for the contribution to 1/T1 from dipolar relaxation by like spins, we could verify that no value of rotational correlation time τc produced a T1 shortening from 3.5 s in CDCl3 to ∼600 ms (see Table 4). Therefore, proton relaxation in propofol must have had contributions from mechanisms other than the intramolecular dipolar interaction, which precluded a simple estimate of τc. At the same time, it is clear that the hard-sphere micellar reorientation times from Table 6 are several orders of magnitude greater than realistic values of the rotational correlation times of either propofol or the surfactants. Therefore, the only conclusions that we can draw are (1) that molecular reorientation of propofol and the surfactants within the micelles is determined by intramicellar degrees of freedom rather than by rigid-sphere micelle reorientation and (2) that proton relaxation in propofol in the surfactant solutions is more complex than a single-mechanism intramolecular homonuclear dipolar relaxation. As discussed above, the room-temperature intermolecular NOEs between Solutol HS15 and propofol were prominently negative. For the system propofol/Tween 80/ D2O-saline, they were negative at 22 °C and weakly positive at 35 °C. The crossover point from the positive to the negative NOE corresponds to the molecular reorientation time of τcrit ) x5/2ω, that is, 0.44 ns for protonproton NOEs at 400 MHz. Again, it is clear that propofol and surfactant molecules undergo a facile mutual reorientation within the swollen micelles, as its time scale is several orders of magnitude smaller than the hard-sphere reorientation time values shown in Table 6. The temperature trend in the system P/T80/D2O-saline is therefore consistent with the shortening of molecular reorientation times as the temperature is increased.

Propofol and Nonionic Surfactants

5. Conclusions In this work we aimed to characterize various physicalchemical factors associated with propofol-surfactant systems. The main conclusions were as follows: (1) Neither the chemical shift nor the T1 of free PEO blocks is resolved from those of the micellar PEO forming the hydrophilic heads of respective surfactants. However, the free PEO manifests itself in diffusion experiments as a fast-diffusing component of the methylene 1H NMR peak (3.7 ppm). (2) Micellar diffusion coefficients (or, in the case of nonmicellar Poloxamer 188, the unimer) were approximately an order of magnitude smaller than those of the fast-diffusing PEO or PPO. (3) The two values were well-resolved in StejskalTanner diffusion plots. (4) Each of the six surfactants solubilizes propofol with the specific solubilization capacity of at least 0.1 g/g in the surfactant concentration range 0-10% (w/v). (5) In solutions of Solutol HS15, Cremophor EL, Tween 80, and Vitamin E TPGS, approximately 1% of the total propofol was partitioned into the free PEO, with the rest residing in surfactant micelles. (6) In solutions of the two Poloxamers, approximately 15% of total propofol was associated with the free PPO, with the rest associated with Poloxamer. (7) In each case, propofol exhibited a single ∆-independent diffusion coefficient which was the weighted average of the two populations, which suggests the presence of fast chemical exchange between them. (8) The stability of solutions at both room temperature and 4 °C and high (>40°) clouding point and gelation temperatures suggests the possibility of their use in pharmaceutical formulations. The practical aim of the study was to investigate novel micellar formulations of propofol potentially appropriate for intravenous administration. Many clinically relevant issues were outside the scope of the present study. These include, among others, susceptibility to bacterial growth; interaction with plasma proteins; hemo- and hepatotoxicity; interaction with cell membranes; pain on injection; and interaction with those metabolites which can be present in the body in high concentrations (such as urea56). (56) Desai, P. R.; Jain, N. J.; Sharma, R. K.; Badahur, P. Colloids Surf., A 2001, 178, 57-69.

Langmuir, Vol. 19, No. 6, 2003 2095

Some nonionic surfactants (such as Triton or Brij) were also outside the scope of this work but would probably be good candidates of a future study. However, at least two clinically relevant issues are noted on the basis of the results presented here. The first is the issue of micellar versus unimer solubilization of the drug. It appears that the surfactant’s facility for unimer solubilization of propofol could be a prerequisite for its intravenous use under some conditions, namely, when the uptake of the drug is slow and the injection results in a dilution of the surfactant to a concentration below its cmc. At least one surfactant, Poloxamer 188, is capable of unimer solubilization of propofol. Significant solubility enhancements of hydrophobic compounds have been noted for other Poloxamers at concentrations below their respective cmc values.11 However, the unimer solubilization capacity of the other surfactants is unclear and requires further investigation. The second issue is that of the pain experienced by patients on intravenous administration of propofol. With a Diprivan emulsion, the pain has been attributed to the free propofol present in blood due to its nonzero solubility in aqueous media. In this respect, the partitioning of propofol into the free ethylene oxide is a negative factor, because free PEO-associated propofol is likely to contribute to the pain experienced by patients. We suggest that the use of purified surfactants, from which the free PEO has been removed chromatographically or otherwise, could provide for a smaller concentration of extramicellar propofol. This will be the subject of further study. Acknowledgment. P.W.K. and K.I.M. acknowledge the support of an ARC-SPIRT grant. The authors thank Mr. Bill Lowe and Dr. W. A. Bubb for technical and NMR spectroscopic assistance, respectively. LA026722G (57) Fro¨mming, K.-H.; Kraus, C.; Mehnert, W. Acta Pharm. Technol. 1990, 36, 214-220. (58) Haque, M. E.; Das, A. R.; Rakshit, A. K.; Moulik, S. P. Langmuir 1996, 12, 4084-4089. (59) Cho, C. H.; Urquidi, J.; Singh, S.; Robinson, G. W. J. Phys. Chem. B 1999, 103, 1991-1994.